Ammonia Removal in Activated Carbons Prepared from Olive Oil Industry Waste

Activated carbons (ACs) from olive stone were prepared using CO2, steam, KOH, and H3PO4 as activating agents. The resultant activated carbons were characterized by proximate and ultimate analysis, N2 adsorption (Brunauer-Emmett-Teller (BET) method), iodine number, Boehm titration, temperature-programmed desorption (TPD), and Fourier transform infrared spectroscopy (FTIR). Ammonia (NH3) was used as a test molecule to be adsorbed. The BET surface areas of the ACs obtained ranged from 1000 to 1986 m2 g-1. Type I isotherms were obtained for all the samples, although steam and H3PO4 ACs showed a significant mesopore contribution. KOH activation resulted in carbon with a high microporosity (98%) and high iodine adsorption (1030 mg g-1). KOH AC prepared with a KOH/pyrolyzed char weight ratio of 2 and at 900 °C showed the highest NH3 adsorption (252 mg g-1), favored by the high microporosity and adequate acidity. Chemical activation (KOH and H3PO4) promotes higher NH3 adsorption than the physical ACs prepared (CO2 and steam). Langmuir and Freundlich adsorption equilibrium models were used to correlate the NH3 adsorption isotherms, obtaining the best fit for the Freundlich equation. The results indicated that olive stone-based activated carbon could be used for commercial AC to remove NH3 from gaseous streams.


Introduction
Activated carbon (AC) is a well-known material employed to separate and remove unwanted substances present in industrial effluents.Because of its affordable price, high porosity, internal surface area, and excellent adsorptive capabilities, AC is the most extensively used adsorbent. 1This material is commonly used in the chemical and food industries, as well as other process industries (drinking water treatment, pharmaceuticals, etc.). 2 Physical and chemical activations are employed to prepare activated carbons (ACs).In the physical method, the raw material is carbonized between 400 and 800 °C, followed by gasification at higher temperatures in an oxidizing gas flow (steam, CO 2 , air, or a combination). 3,4ence, the physical activation is a two-step process but can also be done in a one-step process. 5Carbonization and activation are carried out simultaneously in chemical activation, using a chemical substance (e.g., H 2 SO 4 , H 3 PO 4 , ZnCl 2 , KOH, NaOH) and then subjected to a heat treatment (500-800 °C) in a one-step process. 6,7The surface area, porosity, and adsorption properties of AC are determined by the type of the raw material, as well as the method and process parameters used for carbonization and activation. 8,9nthracite, peat, lignite, bituminous coal, and lignocellulosic materials such as coconut shells and wood are among the most popular carbon-containing feedstocks employed as a precursor to producing AC. 10 In the last decades, an increasing number of investigations related to the production of low-cost AC from agricultural wastes such as sugarcane bagasse, 10 pistachio-nut shells, 11 corn-cob, 3 date stones, 12 neem husk, 13 almond shell, 9 fir bark, 14 jackfruit, 15 cherry stones, 16 coffee grounds, 17 rice husk, 18 peach stones, 19 palm kernel shell, 20 and olive stones, [21][22][23] have been reported.These ACs have displayed good performance and could replace expensive coal-based carbons in various industrial applications to remove different pollutants (e.g., dyes, metal ions, organic pollutants, pharmaceuticals and personal care products).The AC prepared from an abundant waste source would include economic gains and environmental benefits.
In this sense, the olive agribusiness in Chile has about 25,000 hectares of olive trees, with a production of 18,500 tons of oil in 2019 destined for the national and international market, ranking Chile in tenth place among the olive oil-exporting countries in the world. 246][27] However, a large part of this waste is not used and is accumulated, generating environmental problems associated with foul odors. 280][31] Using this process, it is possible to generate add-value products such as liquids and gases with fuel characteristics, and the char can be used as a precursor of the AC. 9 The proposed technological solution aims to strengthen the olive industry in Chile through the integral recovery of its residues through the development of ACs, taking advantage of the intrinsic characteristics of the olive stone.
For this reason, the goal of this study is to evaluate the use of the olive stone to obtain ACs by physical (CO 2 and steam) and chemical activation (KOH and H 3 PO 4 ), testing them for NH 3 adsorption and establishing which method can promote the best AC for the proposed application.Although olive stone as a raw material for obtaining AC has been widely studied, 1,4,8,22,[32][33][34][35][36][37][38][39] few investigations have focused on simultaneously evaluating the four activations carried out in this work.Concerning NH 3 , this gas is one of the most commonly released compounds in the chemical industry, and its presence can be smelt in the air even at low concentrations (< 50 ppm), causing eye, throat, and nose irritation. 40,41The use of AC is a promising approach to removing this gas, considering its simplicity and economic operation. 42Besides, it has been reported that the acidic surface functionalities of AC favor NH 3 adsorption and the direct interaction between the gas molecules and the carbon surface. 43Therefore, the production of AC from the waste of olive oil production can become one of the solutions for the effective elimination of residues and contribute to the sustainability and competitiveness of small and mediumsized industries in this sector.

Materials
Olive stones were supplied by a local olive oil manufacturer (Terramater, Curicó Valley-Region of Maule, Chile).These materials were processed without previous treatment, crushed and sieved until particle sizes between 0.6 and 1 mm were obtained.For comparison purposes, the C Gran and GAC 1240 Norit (Cabot Corporation, Massachusetts, USA) commercial ACs were included in the analysis (see Supplementary Information section).

Preparation of ACs
ACs were prepared using physical and chemical methods.Physical activation was done employing CO 2 (Linde, 99.5%, Santiago, Chile) and steam, and for chemical activation, KOH (Merck, 85% purity, Darmstadt, Germany) and H 3 PO 4 (Merck, 85% purity, Darmstadt, Germany) reagents were used.][46][47][48][49][50] The experimental apparatus used in this research for the preparation of AC is shown in Figure 1.First, olive stones' pyrolysis (carbonization) was carried out in a MELLER PS205 horizontal furnace (New Hampshire, USA) at 400 and 600 °C for 1 h with a heating rate of 5 °C min -1 and a N 2 (Linde, 99.995%, Santiago, Chile) flow rate of 150 mL min -1 .About 40 g of raw material in the alumina ceramic crucible was placed into the furnace.After pyrolysis, the furnace was cooled to room temperature (ca.20 °C) with a N 2 flow.The designations OS-400 and OS-600 refer to olive stones that have been pyrolyzed at 400 and 600 °C, respectively.The pyrolytic char produced at 600 °C was used for ACs prepared in a two-step procedure.

CO 2 activation
The pyrolytic char (4 g) was activated at 850 °C (heating rate of 10 °C min -1 ) with a CO 2 (Linde, 99.999%, Santiago, Chile) flow rate of 100 mL min -1 , keeping the temperature for 1 h.In this case, the sample was obtained in a two-step activation process.For comparison, the raw material (40 g) was directly activated without being previously pyrolyzed, following the one-step activation procedure.The prepared samples were designated as AC-CO 2 1 and AC-CO 2 2, and the numbers 1 or 2 correspond to the activation in one or two-step, respectively.

Steam activation
The previously pyrolyzed olive stone (6.5 g) and the raw material (20 g) were activated in the system indicated above at 900 °C for 1 h, keeping a steam/N 2 (Linde, 99.99%, Santiago, Chile) ratio equal to 4. Next, the samples were heated to the temperature required for activation in N 2 flow (200 mL min -1 ) with a heating rate of 10 °C min -1 .The steam was introduced into the furnace after the temperature was reached.The steam was generated by injecting 0.6 mL min -1 of liquid water into an evaporator by a peristaltic pump (equivalent to 800 mL min -1 of steam determined at 20 °C and 101.32 kPa).After activation, the samples (AC-H 2 O 1 and AC-H 2 O 2) were cooled to room temperature under N 2 flow.

KOH activation
The pyrolyzed char was impregnated with a concentrated KOH solution in the required proportion to obtain two slurries with a KOH/pyrolyzed char weight ratio of 1 or 2. The slurries were heated at 70 °C until incipient dryness.The samples were then dried overnight at 105 °C.After that, the activation was done for 1 h in a N 2 flow of 300 mL min -1 and a heating rate of 5 °C min -1 .Activation was carried out at 800 and 900 °C.Once activated, the first washing was carried out with 0.1 mol L -1 HCl (Merck, 37%, Darmstadt, Germany) solution and later with distilled water until obtaining a pH > 5.The samples were identified as AC-KOH 1-800, AC-KOH 1-900, AC-KOH 2-800, and AC-KOH 2-900, whose numbers 1 or 2 are related to KOH/pyrolyzed weight ratio, and the number 800 or 900 corresponds to the activation temperature.The one-step method was not evaluated for this activation method because the previous test resulted in a low yield (< 11%).It is desirable to produce an AC that has a relatively high yield.
The olive stone and pyrolytic char were mixed with an adequate quantity of a H 3 PO 4 solution to obtain a H 3 PO 4 / sample weight ratio of 1.75.This mixture was stirred at 70 °C until dry and kept on a stove at 105 °C overnight.It was then activated for 1 h at 600 °C (heating rate: 3 °C min -1 ; N 2 flow: 300 mL min -1 ).Finally, the carbon was washed with hot water until pH > 5 and dried at 105 °C to be later stored.The ACs were named AC-H 3 PO 4 1 and AC-H 3 PO 4 2, where the numbers 1 and 2 indicate if the sample was activated in one or two-step.
On completion, the resultant ACs by different methods were weighed to determine their yield.Here, the global yield (Y G ) of char or AC is calculated by dividing the sample weight after activation (W f ) by the weight of the raw material (W i ) using the following equation: (1) Using the pyrolyzed sample as the starting point, an equation was used to figure out the yield (Y p ) of the ACs prepared with the two-step method: Y p (%) = (W p /W i ) × 100 (2)   where W p is the weight of pyrolyzed sample.

Characterization of ACs
Proximate analysis was done following the ASTM D1506 51 and ISO 562 52 standards to determine the content of ash and volatiles, respectively, while the moisture was determined after drying the sample at 105 °C overnight.Ultimate analysis was performed in a Leco CHNS 628 (Michigan, USA) elemental analyzer equipment.
Textural properties were obtained from the N 2 adsorption-desorption isotherm at -196 °C with a Micromeritics 3-Flex instrument (Georgia, USA).The samples (30-50 mg) were pretreated at 300 °C in a vacuum condition for 4 h in the Micromeritics SmartVacPrep equipment (Georgia, USA).The Brunauer-Emmett-Teller (BET) method was used to determine the specific surface area, employing the adsorption branch of the isotherms in the range of 0.02 ≤ P/P 0 ≤ 0.25. 53,54The amount of N 2 adsorbed at P/P 0 = 0.99 corresponds to the total pore volume, including the micro and mesopores. 53All pores in the adsorbent are assumed to be filled with condensed gas at this point.The Dubinin-Radushkevich equation was used to calculate the volume of micropores. 55The volume of mesopores was obtained as the difference between the total volume and the volume of micropores.
The sodium thiosulfate volumetric method was used to determine the iodine number expressed as the milligrams of iodine adsorbed by 1 g of carbon. 56Briefly, the powder AC and a 0.05 mol L -1 iodine solution (Supelco, Reag.Ph Eur, Darmstadt, Germany) are mixed, shaken, and then the filtrate is submitted to titration against a standardized 0.1 mol L -1 Na 2 S 2 O 3 solution (Merck, Darmstadt, Germany).
The surface oxygenated functional groups were determined following the Boehm method. 57The four 0.1 g of samples were mixed with 20 mL of 0.05 mol L -1 of NaOH (Winkler, 98.6%, Santiago, Chile), Na 2 CO 3 (Sigma-Aldrich, 99.5%, Massachusetts, USA), NaHCO 3 (Winkler, > 99%, Santiago, Chile), and HCl (Merck, 37%, Darmstadt, Germany) solutions, respectively.The vials were sealed and stirred for 24 h in a thermal bath at 25 °C.The suspensions are filtered and titrated by 0.05 mol L -1 HCl or 0.05 mol L -1 NaOH solutions.The numbers of the acidic sites were calculated considering that NaOH neutralizes carboxylic, lactonic and phenolic groups; Na 2 CO 3 neutralizes carboxylic and lactonic, and NaHCO 3 neutralizes only carboxylic groups.The amount of HCl that reacted with the AC was used to calculate the number of basic sites.
The chemical nature of the surface was analyzed by temperature-programmed desorption followed by mass spectrometry (TPD-MS), using the same apparatus described in the textural properties test (Micromeritics 3-Flex instrument) coupled with a mass spectrometer (Cirrus 2, MKS Spectra Product, Andover-Massachusetts, USA).The sample (0.05 g) was pre-conditioned in He flow at 45 °C for 1 h before the desorption runs.After that, the sample was heated up to 1000 °C at a heating rate of 10 °C min -1 in He flow (100 mL min -1 ).A quadrupole mass spectrometer was used to detect the amount of CO 2 and CO desorbed as a function of temperature.Calibration was done with calcium oxalate monohydrate for quantification.
The pH of the point of zero charge (pH pzc ) is the pH at which the surface's net charge is zero. 580.1 g of samples were placed in Erlenmeyer flasks with 50 mL of NaCl (Riedel-de Haën, 99.8%, Germany) 0.01 mol L -1 solution.The initial pH (pH i ) was adjusted between 2-12 values using NaOH or HCL solutions of 0.1 mol L -1 and a PL-700PC Gondo instrument pH meter (Taipei, Taiwan).Then, the suspension was shaken for 48 h, and the pH of the filtrate was measured (pH f ).The pH pzc was obtained from the plot of pH f vs. pH i and corresponded to the value where pH f is equal to pH i .
The surface functional groups of carbons were identified using the Fourier transform infrared spectroscopy (FTIR) technique.A Bruker Alpha FT-IR spectrometer was used to record the samples' FTIR spectra between 4000 and 400 cm -1 (Billerica, Massachusetts, USA).First, pellets were prepared by mixing carbon and KBr (1:400 carbon/KBr weight ratio) in an agate mortar.Then, in a hand press, the resulting mixture was compressed.

NH 3 adsorption
Adsorption equilibrium data of NH 3 were carried out in the same equipment employed for the textural properties test (Micromeritics 3-Flex instrument).Before measurement, 50 mg of sample was treated in He flow at 300 °C for 4 h and then exposed to the increasing pressure of anhydrous NH 3 (Indura, 99.5%, Santiago, Chile) up to 90 kPa at room temperature.The adsorption equilibrium amount was calculated at the terminal pressure from the final adsorption amount.
The experimental data were fitted to the Langmuir and Freundlich isotherms. 59The Langmuir isotherm is expressed as: where q c is the adsorbed NH 3 amount per unit mass of solid (cm 3 g -1 ), a m is the maximum adsorption capacity of the solid (mg g -1 ), b is the Langmuir adsorption constant related to the adsorption energy (kPa -1 ), and P is the NH 3 pressure (kPa).The Freundlich isotherm is given by: where K F (cm 3 g -1 kPa -1/n ) and n are the constants, relate to the relative capacity adsorption of adsorbent and the adsorption intensity, respectively. 59,60onlinear regression is used to determine all the parameters from the experimental points.The fitness degree was estimated by the coefficient of determination (R 2 ), and the average relative error (ARE) was calculated by the equation: 61 (5)   where q exp is the experimentally adsorbed amount of NH 3 , q c is the amount of NH 3 calculated using the models, and N is the number of adsorption points.

Starting and pyrolyzed material
Table 1 shows the properties of the raw and pyrolyzed olive stone.The olive stone is characterized by the high volatiles content and carbon and low ash values. 62,63For comparison purposes, the olive stone was pyrolyzed at two temperatures (400 and 600 °C).This sample's proximate and ultimate analysis results are compared with those reported by the references mentioned above, although the moisture can vary between 5 to 10%. 64Sulfurous was not detected in the sample.After pyrolysis of olive stone, the content of volatiles decreased while the carbon increased.It is well known that the increase in temperature promotes the degradation of the constituents of the olive stone (hemicellulose, cellulose and lignin), and these components can be volatilized, and inorganic material (ash) is concentrated. 65As a result of the large quantity of volatiles generated during pyrolysis, yields diminish as the temperature rises from 400 to 600 °C. 11The OS-600 sample was used to produce the AC due to the higher C content compared to OS-400.

Characterization of ACs
The proximate and ultimate analysis results and the yield of the ACs are given in Table 2.The values are similar for all samples and close to the commercial ACs analyzed.The activation increases the carbon content, and the ash tends to concentrate.The %H decreased after the activation treatments, and the nitrogen content remained low, varying between 0.5-1%. 34t is desirable to produce an AC with a relatively high yield, and depending on whether activation was carried out in one-or two-step, this parameter will be affected.When the global yield is considered, the one-and two-step activations report lower yields (20%), except when H 3 PO 4 is used as an activating agent for the one-step method (40%).When CO 2 is used as an activating agent, and the sample is activated in onestep (AC-CO 2 1), a yield of 17% was obtained.The global yield remained in the same order (19%) when the activation was carried out in two steps (AC-CO 2 2 sample).The yield (Y p ) significantly increased up to 70% when calculated from the pyrolyzed olive stone as a starting material.It has been reported that the partial gasification of the raw material or char exposed to CO 2 occurs by the following equation: 8 In this sense, both the effect of a high temperature and CO 2 presence promotes the precursor decomposition and increases the extent of the reaction, 5 causing a decrease in yield (or increased burn-off).In the case of two-step activation, the yield is higher since a large part of the volatiles have already been released.
In steam activation, water reacts with the carbon of the sample to produce carbon monoxide and hydrogen, according to: 8 Therefore, the carbon is consumed, and porosity (micro and mesopores) can be generated.The use of steam or CO 2 makes physical activation more advantageous than chemical activation, requiring more expensive and toxic chemicals and consecutive washing processes to eliminate excess reagents.The global yield obtained for steam activation in one-or two-step was low, registering values between 10 and 11%.During this process, weight loss occurs both due to the evolution of volatile material and the elimination of carbon atoms through steam reaction (equation 6).This low yield agrees with previous research, 66 which indicated that low solid yields (or high burn-off) and high generation of volatiles are associated with steam's capacity to penetrate the solid material, helping the removal and efficient desorption of volatile products.In the case of the AC-H 2 O 2 sample, a yield of 37% was obtained, employing the pyrolyzed sample mass as starting material.This result is because the sample activated with steam was previously pyrolyzed, and a large part of the volatiles was removed.
For chemical activation, specifically the KOH activation, it has been established that this method consists of redox reactions, with the carbon oxidation to CO or CO 2 . 49As shown below, KOH activation involves a sequence of reactions, including dehydration, water gas reaction, and reduction by hydrogen and carbon. 67During these reactions, the biomass is broken down, resulting in the generation of micropores and mesopores structures.
The increase in the KOH/precursor ratio and the temperature causes a slight decrease in the global yield, obtaining 20 and 17% of yield for AC-KOH 1-800 and AC-KOH 2-900 samples, respectively.These yields were 76 and 62% for the previously mentioned samples when calculated from the pyrolyzed material mass, and the effect of the KOH/precursor ratio and temperature on the yield was more evident.Metallic potassium is obtained due to KOH reduction or its transformation products (e.g., K 2 CO 3 ). 50The solid particle breaks down when potassium is intercalated within the graphitic lamellar structure, making granular carbon preparation difficult.This explains why the carbon obtained from this activation turned into a fine powder, and high temperatures or KOH concentration affected the yield.
In addition, the direct activation of the raw olive stone produces a very light material due to the breakdown of the precursor particles after KOH solution impregnation.The attack of hydroxyl ions causes the fragmentation and dissolution of lignin and hemicellulose. 39It explains the meager yield obtained (< 11%) in the preliminary test using unpyrolyzed olive stone and why it was not used in the AC preparation.
When H 3 PO 4 was used as an activating agent, a Y G of 40 was obtained for the AC-H 3 PO 4 1 sample, while the AC-H 3 PO 4 2 sample registered a Y G of 18% (68%, considering the pyrolyzed material).This trend is similar to that observed for the evaluated physical activations.As mentioned before, the sample impregnation generates reactions that promote cellulose fragmentation and other structural components of the olive stone, so the raw material is more affected than the pyrolyzed char.Because the acid separates the cellulose fibers and causes partial depolymerization of hemicellulose and lignin, the particle treated with H 3 PO 4 becomes elastic, causing a decrease in mechanical resistance and particle swelling.Similarly, because many tars have been detected on the particle's surface, impregnation promotes carbon conversion. 45These tars come from acid-catalyzed cellulose depolymerization, dehydration, and condensation, resulting in aromatic and reactive compounds with some cross-linking.Figure 2 depicts the nitrogen adsorption-desorption isotherms for the physically and chemically activated samples, while Table 3 reports the textural properties obtained from the adsorption isotherms.All the carbons presented type I isotherms, characteristic of predominant microporous solids.For samples activated with CO 2 and KOH, a percentage of micropores (%V mi /V T ) higher than 90% was obtained, while the carbons obtained by steam and H 3 PO 4 activation showed a micro-mesoporous structure. 8,38,67,68or micro-mesoporous carbons, an H4-type hysteresis loop was observed, usually found in solids that consist of aggregates or agglomerates of particles with slit-shaped pores. 3These samples showed percentages of micropores between 60-78%, with an average pore diameter very close to 2 nm, indicative that small mesopores are present.It has been reported that CO 2 activation favors the creation of microporosity while the use of steam increases microporosity at the beginning of the activation, which are then transformed into mesopores, obtaining carbons with a lower percentage micropore than CO 2 -activated samples. 35,69he chemically activated samples showed the highest surface areas, highlighting the AC-KOH 2-900 (1823 m 2 g -1 ) and AC-H 3 PO 4 1 (1986 m 2 g -1 ) samples.For the KOHactivated samples, increasing the KOH/precursor ratio (from 1 to 2) and the activation temperature (from 800 to 900 °C) improves the surface area.This result can be expected because more KOH is intercalated into the material structure, favoring the porosity generation during impregnation and subsequent activation. 38,70It has been reported the presence of mesopores when wood carbons were activated with H 3 PO 4 , while the carbons obtained by KOH activation reported a microporous structure. 71he one-step sample activated with CO 2 (AC-CO 2 1) has a surface area of 1011 m 2 g -1 , a higher value than the sample activated in a two-step (AC-CO 2 2, 810 m 2 g -1 ).However, the percentage of micropores did not change (ca.90%).This same behavior was reported for H 3 PO 4 -

Sample
S BET / (m 2 g -1 ) V mi / (cm 3 g -1 ) V me / (cm 3 g -1 ) V tot / (cm activated carbons, where the surface area of AC-H 3 PO 4 1 was higher than AC-H 3 PO 4 2. The opposite trend was observed for the steam-activated sample, whose surface area of the AC-H 2 O 1 sample (597 m 2 g -1 ) was lower than that activated in a two-step (AC-H 2 O 2, 1511 m 2 g -1 ).These results indicate the importance of the initial contact surface with the activating agent and the temperature, activation time, activating agent flow, and heating ramp that influence the diffusion and removal of gaseous products from the surface of the coal. 4,5,34t is well known that the non-carbonized substance interacts with the activating agent more intensely than the carbonized material. 38Hence, carbons obtained in one-step activation show higher BET surface area (S BET ) than carbons prepared in two-step activation (using OS-600 sample for the steam activation).However, this behavior was not observed for steam-activated samples.It is possible that the one-step activation conditions were very aggressive, consuming the sample and the destruction of pores.This hypothesis is established considering the low global yield obtained for the two activated carbons (AC-H 2 O 1 and AC-H 2 O 2) and other investigations, 2,3 indicating that the external burning of carbon particles affects the porosity.For carbons prepared in the two-step activation, the sample carbonized in nitrogen has a more organized structure and less developed porosity than carbonized in steam.This fact is related to a lower reaction rate for the nitrogencarbonized sample. 72he iodine number allows one to quantify the AC's adsorptive capacity and can be used to determine pores with dimensions ≥ 1.0 nm. 44Iodine number of the samples with the highest surface area by each method and commercial carbons is reported in Table 4.Chemical KOH activation generates a carbon with high iodine adsorption capacity, with an iodine number of 1030 mg g -1 for AC-KOH 2-900 sample, even higher than those obtained for commercial carbons.Considering that iodine adsorption is indicative of the adsorption capacity in micropores, 73 this result is well-correlated with its textural properties (Table 3) because a high percentage of micropores was reported (98%).The other carbons showed a lower iodine number than the KOH sample.It may be due to the creation of meso and macropore structures, thus decreasing the adsorptive capacity of iodine. 2 Similar values of iodine number for carbons based on olive stone have been reported by other researchers. 29,36,74oehm titration results of selected carbons are shown in Table 4.The chemistry of the surface and the porosity have a significant impact on the performance of ACs.Acidic groups include carbonyl, carboxyl, phenolic hydroxyl, and lactonic groups, whereas basic groups include pyrene, chromene, quinone, and the graphene structure's electrons. 75The AC surface's acid/base nature is determined by the groups and delocalized electrons of the graphitic structure. 76Depending on the activation method and preparation conditions, functional groups on the carbon surface with a wide range of concentrations are favored. 74cidic surface groups decompose at high temperatures, implying that high activation temperatures favor low acidity. 77Because of this, physically activated carbons (like CO 2 and steam) tend to be basic, which is in line with the pH pzc values (7.8 and 8.2, respectively) found in Table 4 and shown in Figure S1 (SI section).Therefore, the surface groups in the AC-CO 2 1 and AC-H 2 O 2 samples are mostly basic.The same trend was registered for the AC-KOH 2-900 sample.In this sense, carbons prepared from hazelnut bagasse and activated with KOH (KOH/precursor mass ratio between 1-3) have predominantly basic surface groups. 6The sample activated with KOH registered the highest amount of carboxylic groups, with total values of the basic groups slightly higher than acidic groups.This result promotes a little acidic pH pzc value (6.7).
On the other hand, treatment with H 3 PO 4 generates an activated carbon of an acidic nature (pH pzc = 3.2), with fewer carboxylic groups than the sample activated with KOH.However, this sample recorded the highest content of phenolic groups.These results are consistent with those registered in the TPD tests, obtaining a more remarkable CO 2 evolution in the sample activated with KOH.In contrast, the one activated with H 3 PO 4 showed a higher evolution of CO.
For AC-H 3 PO 4 1 sample, a higher proportion of acid groups was generated, a result also obtained by other researchers. 58,74,780][81] The commercial activated carbons Norit GAC 1240 and C Gran showed pH pzc measurements related to their neutral and acid nature. 82he total capacity of oxide groups increased using H 3 PO 4 as an activating agent.The generation by surface oxidation of acidic functional groups containing oxygen and/or phosphorous and its attachment to the surface while the porosity is being developed has been proposed. 83Even in the case of ACs prepared using H 3 PO 4 , the carboxylic group is thought to be the only strong acidic group detectable by NaHCO 3 . 69,84,85The -OH group linked to phosphoruscontaining acids adsorbed on the surface has been partly responsible for the surface's strong acidity (i.e., -PO 2 H 2 ). 86ased on the investigation mentioned before, it is suggested that the acidic groups detected should be considered as both oxygen-containing and oxygen/phosphorus acidic groups.
The surface chemistry of the ACs was evaluated using temperature-programmed desorption (TPD) experiments coupled to mass.The temperature at which CO and CO 2 are evolved during TPD can characterize the functional groups that contain oxygen.The CO 2 and CO desorption profiles obtained by TPD-MS are shown in Figure 3, while the amounts desorbed are compiled in Table 5.It has been proposed as a general trend that CO 2 evolves at temperatures between 300 and 800 °C and corresponds to the degradation of the surface oxygen groups that are more acidic and less stable.At temperatures near 300 °C, carboxylic groups degrade to CO 2 , followed by lactones and anhydrides (which give both CO and CO 2 ) at temperatures above 800 °C.[89][90][91] AC-KOH 2-900 sample (Figure 3a) showed the highest value for CO 2 evolution (3.306 mmol g -1 ), which can be explained by the high degree of more acidic surface groups, such as carboxylic groups.The presence of this acidic group was also corroborated by the Boehm and FTIR characterization.As will be discussed later, the existence of this surface oxygen group plays an essential role in NH 3 adsorption.For AC-H 3 PO 4 1 sample, the CO 2 TPD profile shows the majority presence of lactones and/or anhydrides groups.At the same time, the region related to the decomposition of carboxylic groups did not have a significant contribution to the profile.This observation could corroborate the hypothesis raised concerning the overestimation of carboxylic groups in the Boehm treatment.
The CO desorption profiles are shown in Figure 3b.AC-H 3 PO 4 1 sample had the highest CO evolution (5.170 mmol g -1 ), registering broadband between 350 and 1000 °C with a maximum at 848 °C and a shoulder near  686 °C.The signal at 686 °C is attributed to phenol and hydroquinones, while the maximum at 848 °C is assigned to carbonyl and quinones. 92he difference in the oxygenated surface groups was noticed depending on the chemical agent used for the activation.For KOH activated sample (AC-KOH 2-900), oxygenated surface groups degrade primarily as CO 2 (anhydrides, lactones, carboxylic acid, etc.) during TPD tests, while for H 3 PO 4 activated sample (AC-H 3 PO 4 2), those groups decompose as CO (carbonyl, semiquinone, etc.) are the most numerous. 38In this sense, the oxygenphosphorous bond in the C-O-P system is the weakest in H 3 PO 4 activated samples.Therefore, this bond is broken at high temperatures, leaving only an O atom linked to a carbon site and favoring the CO desorption. 93TIR spectra of the obtained ACs are shown in Figure 4.The assignments of various bands observed are listed in Table 6.The presence of the oxygen functional groups was determined by the FTIR, which agrees with the Boehm results.The samples' spectra were very similar, and an intense band at 3428 cm -1 for AC-H 2 O 2 sample can be noticed.Considering that this sample was activated using steam and the band can also be assigned to adsorbed water, 94 the contribution of water on the spectrum is not ruled out.
It has been reported in H 3 PO 4 ACs prepared from woody biomass, a broadband between 900-1300 cm -1 with a maximum at 1116-1142 cm -1 .This band is related to oxygen and phosphorous compounds in the ACs. 74In our results, considering the similarity of the carbons' spectra that were prepared using different methods and the overlap of the absorption bands, an unambiguous assignment is complex, and just the presence of phosphorous compounds can be suggested.
For Norit C Gran commercial AC, the intense band at 1582 cm -1 indicates a more non-polar (hydrophobic) surface than the other prepared carbons. 15 3 adsorption NH 3 adsorption isotherms of the selected samples at 20 °C are plotted in Figure 5.The maximum amount of NH 3 adsorbed was determined from these isotherms and is reported in Table 7.The volume adsorbed indicates a high adsorption capacity for all the ACs prepared compared to other carbons discussed in the literature and the commercial ones used as references.For example, AC-KOH 2-900 AC reports the higher NH 3 volume adsorbed (332.9 cm 3 g -1 ), corresponding to 252 mg g -1 .The high adsorption capacity obtained for the samples could be due to the oxygen functional groups, especially the carboxylic groups.100 Figure 6 depicts the interaction between ammonia molecule and the carboxylic groups of ACs.
Boehm titration, FTIR and TPD results have demonstrated the presence of acidic functional groups (phenolic, lactonic and carboxylic) that can explain the high NH 3 adsorption capacity obtained.It has been found that an increase in the number of functional oxygen groups

Table 6. Assignments of infrared bands observed in the ACs prepared
Band position / cm -1 Assignment 3428 OH stretches in hydroxyl, carboxylic and phenolic groups 15,94 2924, 2835 C-H stretching modes in aliphatic 15,95 1732, 1705, 1637 C=O and C-O stretching of carboxylic acid, phenolic ester and conjugated ketonic structures 95,96 1582 C=C stretching in aromatic rings 97,98 1460 C-O stretching vibration from the carboxyl group 15,94 1371 asymmetric bending vibration of the -CH 3 group 15,94 1173 stretch vibration of C-C, C-O or C-H from the carboxyl groups (-COOH) 94 1105, 1058 C-O group in carboxylic and alcoholic groups 70,99 Below 1000 aromatic, out of plane C-H bending with different substitution degrees 95 on the surface of modified carbons favors NH 3 adsorption capacity and is related to the acid site type present.Due to the electronegative nature of nitrogen, the hydrogen atoms of the NH 3 molecule interact with the oxygen atoms linked as hydroxyl (H-O) and carbonyl (C-O) groups on the surface because of the electrostatic forces (see Figure 6). 105n this sense, TPD results of the AC-KOH 2-900 sample indicated the presence of a higher proportion of more acidic and less stable surface groups than other samples and explained the high NH 3 adsorption capacity obtained. 90This assertion is established by correlating the adsorption capacity of NH 3 with the amount of more acidic surface groups (CO 2 ) that was released during the TPD test (Figure 7a).A linear trend was observed (R 2 = 0.857), although the increase in the NH 3 adsorption capacity with the more acidic surface oxygen groups was clearer for the AC-KOH 2-900 sample.In this sense, Gonçalves et al. 87 showed a linear correlation between the quantity of the more acidic oxygen surface groups and the NH 3 adsorption capacity in a commercial carbon (MAST).
In order to evaluate the influence of microporosity on the adsorption capacity of NH 3 , these two parameters were plotted and are shown in Figure 7b.There is no relationship between the parameters that can be corroborated with the behavior of AC-KOH 2-900 and AC-H 3 PO 4 1 samples with an equal volume of micropores (ca.0.73 cm 3 g -1 ) but very different NH 3 adsorption (252 vs. 156 mg g -1 , respectively).This tendency is identical if it is plotted the NH 3 adsorption against the surface area and mesopore volume.This behavior was also reported by Yeom et al. 102 in a commercial AC, who indicated that the NH 3 adsorption capacity was unaffected by the surface area.However, other authors have established the preference for adsorbents with large surface areas and small pore sizes for NH 3 removal. 101,106Additionally, mesoporous samples have been proposed to adsorb ammonia because ordered and regular porous structures with well-defined channels favor the adsorption of NH 3 . 59,102These different results suggest that the nature and preparation conditions  Regarding the adsorption models, the experimental data were adjusted to the Langmuir and Freundlich adsorption isotherm models.The Langmuir model implies monolayer adsorption on a homogeneous surface, with all active sites having the same energy and being equivalent.In addition, the Langmuir model considers dynamic equilibrium and no adsorbate interaction. 107The Freundlich model is based on the following assumptions: (i) multilayer adsorption on heterogeneous surfaces, (ii) adsorbate interactions and (iii) increase in the adsorption capacity with the concentration of the analyte. 14ccording to the results in Table 8, the lower ARE and the closest R 2 values to 1 indicate a good correlation for the Freundlich adsorption isotherm related to multilayer adsorption on a heterogeneous surface. 14Thus, the Freundlich model best describes NH 3 adsorption on prepared and reference ACs.In this sense, several authors have reported the same behavior in the fitting of the experimental data to the Freundlich model, 59,108,109 and the value of n > 1 indicates favorable adsorption conditions. 67

Conclusions
In this study, ACs using olive stones were prepared.Depending on the activation method, the surface area of carbons ranges between 1000-1986 m 2 g -1 , indicating that olive stone was successfully activated.Type I isotherms were observed in all the samples, characteristics of microporous solids; however, the isotherms obtained for steam and H 3 PO 4 activated samples indicated the contribution of mesopores.KOH-AC reported the highest  iodine adsorptive capacity (1030 mg g -1 ), which correlates with the high percentage of microporous obtained (98%).Surface area results showed the importance of the initial contact surface with the activating agent depending on whether the activation is in one or two steps, affecting the diffusion and removal of gaseous products from the sample.The presence of acidic functional groups (carboxylic, lactonic, phenolic) was observed by Boehm titration, FTIR, and TPD characterizations.It can explain the NH 3 adsorption capacity obtained, especially for AC-KOH 2-900 sample (252 mg NH 3 g carbon -1 ).The TPD profile of KOH-AC showed the highest CO 2 evolution because more acidic surface groups were reported (carboxylic groups), and a linear trend was observed between the amount of NH 3 adsorbed and these surface groups.However, the surface area and micropore volume of samples did not affect the adsorption capacity of NH 3 .The NH 3 adsorption amounts of the physical and chemical prepared ACs are in the following order: AC-KOH 2-900 > AC-H 3 PO 4 1 > AC-CO 2 1 > AC-H 2 O 2. The good performance for the NH 3 adsorption obtained for the KOH-activated sample could be explained by the combination of the high microporosity and adequate surface acidity.The Freundlich model best fit the experimental data, implying that the multilayer adsorption on a heterogeneous surface was dominant.

Figure 1 .
Figure 1.Schematic diagram of the pyrolysis/activation experimental system.

Table 1 .
Properties of raw and pyrolyzed olive stone a Dry basis.Y G : global yield.C + H 2 O → CO + H 2

Table 2 .
Proximate and ultimate analysis and yield of ACs a Dry basis; b commercial AC.Values in parenthesis correspond to yield Y p ; Y G : global yield.

Table 4 .
Iodine adsorption and Boehm titration of selected carbons

Table 5 .
Amounts of CO and CO 2 desorbed measured by TPD-MS

Table 7 .
NH 3 adsorption capacity of some adsorbents a Activated carbon prepared in this work; b commercial activated carbon.S BET : BET surface area.Figure 6. Interaction of ammonia with the carboxylic groups of AC.

Table 8 .
Fit of the experimental data of NH 3 adsorption at 20 °C to the Langmuir and Freundlich isotherms m : solid's maximum adsorption capacity; b: Langmuir adsorption constant related to the adsorption energy; ARE: average relative error; R 2 : coefficient of determination; K F : constant relate to the relative capacity adsorption of adsorbent; n: constant of adsorption intensity.